Google DeepMind published research demonstrating that AI agents can accelerate stress-testing of cryptographic systems, identifying two major weaknesses in the traditional 1024-bit RSA encryption scheme. The findings were achieved under human guidance, with DeepMind's models uncovering vulnerabilities that challenge historical assumptions about computational difficulty. Modern 2048-bit RSA systems remain unaffected by these specific vulnerabilities. However, DeepMind emphasized that as AI-driven mathematics and computation advance, cryptographic standards must undergo continuous stress-testing to ensure long-term security against evolving computational capabilities.